5 Major Announcements In OpenAI’s DevDay 2023

just finished watching the 45-minute keynote of OpenAI’s first ever Dev Day, and damn, it was packed with exciting announcements! It would have been awesome to be there in person, since they gave out $500 OpenAI credits to those who attended in person.

I took notes while watching, and these are the top five product announcements that I’m most excited about:

1. New GPT-4 Turbo
2. Assistants API
3. Customizable GPTs
4. New Text-To-Speech model
5. GPT App Store (Coming soon)

Let’s unpack each of these products:

New GPT-4 Turbo

The new GPT-4 Turbo is a more capable version of the original GPT-4.

• It’s cheaper than the previous GPT-4 model
• Supports a 128K context window
• Has a knowledge cut-off until April 2023!
• It has improved function calling

Check out the price comparison of the GPT-4 Turbo with the current GPT-4.

It is 3x cheaper for the input token and 2x cheaper for the output token.

Also, the 128k context window is huge—large enough to fit the equivalent of more than 300 pages of text in a single prompt!

Assistants API

The AI Assistants API allows developers to build AI assistants within their own applications. The Assistants API currently supports three types of tools:

• Code Interpreter
• Retrieval
• Function calling

• Thread: a conversation session between an Assistant and the user.
• Messages: messages can include text, images, and other files. Messages are stored as a list on the Thread.
• Run: an invocation of an Assistant on a Thread. The Assistant uses it’s configuration and the Thread’s Messages to perform tasks by calling models and tools.
• Run Step: a detailed list of steps the Assistant took as part of a Run. An Assistant can call tools or create Messages during it’s run.

If you’re tech-savvy, you’d notice that this works similarly with AgentGPT or LangChain.

Customizable GPTs

GPTs allow anyone to create a custom version of ChatGPT to be more helpful in their daily life, at specific tasks, at work, or at home.

One example that Sam Altman showed during his keynote was a Canva GPT that generates two design posters from a single text prompt. Clicking on the design redirects you to the Canva website for further modifications. How cool is that?

GPT App Store (Coming soon)

Connected to the GPTs, OpenAI will launch a GPT Store where developers can upload their custom GPTs. Similar to the App Store or Google Play Store, the most useful GPTs developed by users will get spotlighted in different categories like productivity, education, and “just for fun”.

Also, in the coming months, developers will be able to earn money based on the usage of their GPTs. This is a new way for app developers to monetize.

New Text-To-Speech model

OpenAI also released an API that converts text to speech:

• Text-to-speech 1: The latest text-to-speech model, optimized for speed
• Text-to-speech 1 HD: The latest text-to-speech model, optimized for quality

Check out some examples of voice options here.

Unfortunately, you can’t replicate your own voice just yet.

There were some other announcements not included in this list:

• A new gpt-3.5-turbo-1106 with longer context window and cheaper API access
• Release of Dall-E 3 API
• The model picker in ChatGPT was removed—you can now access Dall-E, browsing, and data analysis all without switching models.
• Multimodality: GPT-4, Dall-E 3, and Text-To-Speech can now work together in your application

With an OpenAI account, you can now try some of the new products like the Assistant.

Final Thoughts

After watching the highlights from OpenAI’s DevDay 2023, I am thrilled by the rapid pace of innovation. A lot of developers are probably working to update their apps with the latest AI models—I will be working on updating my own product too!

The only downside is that startups with similar offerings will have a hard time competing with OpenAI’s relentless improvement cycle. But overall, this is a huge leap forward for developers, and I can’t wait to start building.